IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, … · 562 IEEE TRANSACTIONS ON BIOMEDICAL...

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, NO. 2, FEBRUARY 2015 561

In Vivo Irreversible Electroporation Kidney Ablation:Experimentally Correlated Numerical Models

Robert E. Neal, II∗, Member, IEEE, Paulo A. Garcia, Member, IEEE, Helen Kavnoudias, Franklin Rosenfeldt,Catriona A. Mclean, Victoria Earl, Joanne Bergman, Rafael V. Davalos, Member, IEEE, and Kenneth R. Thomson

Abstract—Irreversible electroporation (IRE) ablation uses briefelectric pulses to kill a volume of tissue without damaging thestructures contraindicated for surgical resection or thermal ab-lation, including blood vessels and ureters. IRE offers a targetednephron-sparing approach for treating kidney tumors, but the rel-evant organ-specific electrical properties and cellular susceptibilityto IRE electric pulses remain to be characterized. Here, a pulse pro-tocol of 100 electric pulses, each 100 μs long, is delivered at 1 pulse/sto canine kidneys at three different voltage-to-distance ratios whilemeasuring intrapulse current, completed 6 h before humane eu-thanasia. Numerical models were correlated with lesions and elec-trical measurements to determine electrical conductivity behaviorand lethal electric field threshold. Three methods for modeling tis-sue response to the pulses were investigated (static, linear dynamic,and asymmetrical sigmoid dynamic), where the asymmetrical sig-moid dynamic conductivity function most accurately and preciselymatched lesion dimensions, with a lethal electric field threshold of575 ± 67 V/cm for the protocols used. The linear dynamic modelalso attains accurate predictions with a simpler function. Thesefindings can aid renal IRE treatment planning under varying elec-trode geometries and pulse strengths. Histology showed a whollynecrotic core lesion at the highest electric fields, surrounded by atransitional perimeter of differential tissue viability dependent onrenal structure.

Index Terms—Bioimpedance, dynamic conductivity, IRE, non-thermal focal tumor ablation, translational targeted cancertherapy.

I. INTRODUCTION

INCREASED diagnostic imaging has augmented new renalcancer diagnoses, with over half of cases now detectedManuscript received August 16, 2013; revised August 19, 2014; accepted

August 30, 2014. Date of publication September 25, 2014; date of currentversion January 16, 2015. This work was supported by The Flack Trustees,ICTAS Multi-scale Bio-Engineered Devices Center, and Whitaker Programs.Modeling supported by NSF CAREER CBET-1055913. R. E. Neal, II and P.A. Garcia contributed equally to this work. Asterisk indicates correspondingauthor.

∗R. E. Neal, II, is with the Department of Radiology, The Alfred Hospital,Melbourne, VIC 3004, Australia (e-mail: [email protected]).

P. A. Garcia and R. V. Davalos are with the BioElectroMechanical SystemsLaboratory in the Virginia Tech-Wake Forest School of Biomedical Engineeringand Sciences, Virginia Tech, Blacksburg, VA 24061 USA (e-mail: [email protected]; [email protected]).

H. Kavnoudias, V. Earl, J. Bergmann, and K. R. Thomson are withthe Department of Radiology, The Alfred Hospital, Melbourne, VIC 3004,Australia (e-mail: [email protected]; [email protected]; [email protected]; [email protected]).

F. Rosenfeldt is with the Department of Surgery, The Alfred Hospital, Mel-bourne, VIC 3004, Australia (e-mail: [email protected]).

C. A. Mclean is with the Department of Anatomical Pathology, The AlfredHospital, Melbourne, VIC 3004, Australia (e-mail: [email protected]).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TBME.2014.2360374

incidentally, including nonaggressive tumors and renal cell car-cinomas with low incidence of metastatic progression [1]. Inresponse, nephron-sparing approaches are being used morecommonly to eliminate abnormal masses while conserving renalfunction in the healthy parenchyma [2]; particularly for patientswith comorbidities, multiple renal carcinoma nodules, or ge-netic predisposition for recurrent tumors. Such techniques, in-cluding partial nephrectomies and thermal ablation [3], achievecomparable medium-term outcomes to radical nephrectomy [4].Thermal techniques are often limited to tumors �3 cm in non-central regions due to blood perfusion thermal sink and potentialcollecting system damage [3], [5].

Irreversible electroporation (IRE) ablation uses a series ofbrief (∼100 μs) but intense (1–3 kV) electric pulses delivered toa targeted tissue volume via needle electrodes [6], [7], alteringthe resting cell transmembrane potentials to create irrecover-able nanoscale defects [8]. IRE ablation is unaffected by bloodperfusion thermal sinks and does not affect acellular tissue con-stituents, preserving patency of the major vasculature, ductalsystems, and other sensitive structures [9], [10]. Treatments arecontrolled by electrode placement and pulse parameters, andcan be monitored in real time [11]–[13]. Clinical trials are eval-uating IRE safety and efficacy in organs including kidney, liver,lung, prostate, and pancreas [14]–[20].

To ensure complete tumor destruction while sparing remain-ing healthy tissue, IRE protocols require accurate predictivetreatment planning models. IRE effects change with clinical in-dication and organ [20], likely due changes in cell and tissuecomposition. Preclinical renal IRE demonstrates sparing of col-lecting system, renal calices, medulla pelvis, and vasculaturewithin ablated regions [21]–[24]. However, characterization ofrenal tissue electrical properties and effective lethal IRE electricfield threshold remains to be determined. These differences maycause discrepancy between simulation predictions and in vivoablations [25].

Unique renal anatomy will affect electric field distributionand cellular sensitivity to the electric pulses. These propertiesare also dynamic, as aqueous pathways through cell membranesfrom electroporation (EP) dramatically change local tissue elec-trical conductivity [26], thus altering the electric field and abla-tion zone [27]–[29]. Ex vivo porcine renal experiments matchedthis behavior with an asymmetrical sigmoid function of localelectric field σ(|E|) [30]. Other studies examining dynamic con-ductivity behavior suggest its prevalence and effect on electricfield distributions. In vivo rat liver experiments showed a 3.8×increase in conductivity [31]. Several tissues were modeled witha symmetrical sigmoid conductivity function in [28] and [29].

0018-9294 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications standards/publications/rights/index.html for more information.

562 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, NO. 2, FEBRUARY 2015

Calibrated canine brain ablations used a σ(|E|) equation basedon EP and thermal tissue change values in the literature [32].These dynamic conductivity functions better match experimen-tal electrical currents than static models [33]. However, it re-mains important to characterize the tissue-specific properties forthese functions and confirm which dynamic conductivity elec-tric field distributions appropriately match lesion shapes underIRE tissue ablation protocols.

II. METHODS

Here, we expand upon the findings in [34] by comparing nu-merical simulations using three conductivity functions to rep-resent three unique electrode and voltage strength protocols. Inthis study, in vivo canine renal ablations are performed whilepre- and intrapulse electrical current measurements are taken todetermine conductivity parameters. Lesions are correlated withone static and two dynamic numerical models to determine aneffective electric field threshold, EIRE , for renal tissue using100 pulses, each 100-μs-long, and delivered at 1 pulse/s.

A. In Vivo Renal Experiments

1) Animal Preparation: IRE ablations were performed incanine kidneys in a procedure approved by the AMREP Ani-mal Ethics Committee. Four previously condemned male grey-hounds weighing approximately 30 kg were procured fromsources approved by the animal ethics committee for use in re-search. All canines were quarantined �24 h prior to proceduresfor acclimation and to ensure health. Subjects were premedi-cated with acetylpromazine (0.1 mg/kg), atropine (0.05 mg/kg),and morphine (0.2 mg/kg) prior to general anesthesia inducedwith propofol (6 mg/kg, then 0.5 mg/kg/min) and maintainedwith inhaled isofluorane (1–2%). Anesthetic depth was mon-itored with EEG brain activity bispectral index (Covidien,Dublin, Ireland). After ensuring adequate anesthesia, a midlineincision was made and mesenchymal tissue was maneuvered toaccess the kidney. Immediately prior to pulse delivery, pancuro-nium (initial 0.2 mg/kg dose, adjusted based on contractions)was delivered intravenously to mitigate pulse-induced musclecontraction.

2) Experimental Procedure: Two parallel 18 gauge needleelectrodes (1.0 mm diameter and 1.0 cm in exposure) were in-serted as pairs into the superior, middle, or inferior lobe of thekidney at a separation distance of either 1.0 or 1.5 cm (center-to-center). A BTX ECM830 pulse generator (Harvard Apparatus,Cambridge, MA) was used to deliver an initial 100 μs prepulseof voltage corresponding to a 50 V/cm voltage-to-distance ratiobetween the electrodes, with distance being the center-to-centerseparation. The prepulse is used to determine baseline electricalconductivity for the numerical models. Electrical current wasmeasured with a Tektronix TCP305 current probe connected toa TCPA300 amplifier (both Tektronix, Beaverton, OR). A ProtekDS0-2090 USB computer-interface oscilloscope provided cur-rent measurements on a laptop using the included DSO-2090software (both GS Instruments, Incheon, Korea). A schematicof the experimental setup can be found in Fig. 1(a).

Fig. 1. Experimental setup and sample current waveforms. (a) Experimentalequipment setup schematic, demonstrating approximate location of electrodeinsertions for horizontal (“H”) and vertical (“V”) arrangements. (b) Dimensionlabeling conventions. (c) 50 V prepulse electrical current at 1 cm separation,grid = 0.25 A. The lack of rise in intrapulse conductivity suggests no significantmembrane EP during prepulse delivery. (d) Electrical current for pulses 40–50of 1750 V at 1 cm separation, grid = 5 A. Intrapulse current rise suggestscontinued conductivity increase and EP.

TABLE IIn Vivo CANINE RENAL EXPERIMENT PROTOCOLS

Setup Separation, cm Voltage, V Voltage–distance ratio, V/cm

1 1 1250 12502 1 1750 17503 1.5 2250 1500

Following the prepulse, a series of 100 pulses, each 100 μslong, at a rate of 1 pulse/s was delivered, reversing polarity after50 pulses. A 5 s pause was encountered after pulses 10 and50 to store the data. The three protocol applied voltages andelectrode separation distances (center-to-center) are shown inTable I. The electric field from needle electrodes has an inho-mogeneous, nonlinear distribution, which varies spatially. and isthus highly spatially variable. Therefore, the voltage-to-distanceratio is provided as a general reference metric for pulse inten-sity based on needle separation and voltage. Electrodes wereremoved following pulse delivery. Ablations were performedin the remaining lobes before repeating the procedure on thecontralateral kidney, resulting in a total of three ablations perkidney and six per canine.

Trials were repeated until satisfying n � 4 trials per setup andattaining ablation height and width dimensions with standarddeviations less than 10% of their average for each protocol.These criteria were considered acceptably reliable data withthe minimum required subject number. Fourteen ablations wereperformed in eight kidneys (see Table I).

3) Lesion Determination: Animals were maintained underanesthesia 6 h following electric pulse delivery, as this hasbeen noted a sufficient period to allow lesion development inother tissues [35], [36], before euthanasia via pentobarbital over-dose. Kidneys were removed; trial sections were separated and

NEAL, II, et al.: IN VIVO IRREVERSIBLE ELECTROPORATION KIDNEY ABLATION: EXPERIMENTALLY CORRELATED NUMERICAL MODELS 563

preserved in 10% neutral buffered formalin for 48 h prior to sec-tioning into 5 mm slices and photographed. Images were blindedand analyzed with imageJ (National Institutes for Health, MD)to determine maximum lesion length and width dimensions [seeFig. 1(b)], with statistical significance tested via ANOVA usingMicrosoft Excel (Microsoft, Redmond, WA).

4) Histology and Viability Evaluation: Formalin-only pre-served samples were sectioned and stained with haemotoxylinand eosin. Triphenyltetrazoliumchloride (TTC) dye was used inearly trials to examine gross cell viability. In these trials, har-vested regions were sectioned into 5 mm slices and immersed in5% TTC phosphate buffered saline for 20 min prior to movingto 10% buffered formalin.

B. Numerical Modeling

Numerical models offer a platform for predicting EP effectsby simulating distributions of electrical conductivity, electricfield, and temperature. The electric field distribution combinedwith an effective lethal electric field threshold for IRE at a givenset of secondary pulse parameters, EIRE , enables predictingablation volumes under different electrode arrangements andapplied voltages. This ability is vital for optimizing treatmentprotocols for patient-specific scenarios. While these effects havebeen examined in other tissues and cells in suspension [37],accurate model predictions must be standardized to in vivo data.Thus, here we examine and calibrate numerical simulations toenable treatment planning under three experimental protocolsusing three tissue electrical conductivity behaviors.

1) Kidney Segmentation and 3-D Reconstruction: The com-putational model domain was constructed from donated MRIscans of a similarly sized canine, scaled by 1.21 times to matchthe experimental kidneys. Mimics 14.1 image analysis software(Materialise, Leuven, BG) was used to segment axial kidneyMRI traces, which were integrated into a 3-D kidney volume forexport to 3-matic version 6.1 (Materialise, Leuven, BG) to gen-erate a volumetric mesh compatible with Comsol Multiphysicsfinite-element modeling software (Comsol Multiphysics, v.4.2a,Stockholm, Sweden).

Electrodes were simulated as cylinders, each 1 cm long and1 mm in diameter, and separated by 1 or 1.5 cm (center-to-center)to represent the experiments. Pairs were inserted into the 3-Dkidney mesh with tips 1.5 cm deep, thus leaving 5 mm betweenthe proximal electrode edge and kidney surface. Because theresults of Neal et al. [34] showed negligible effects from themodeled electrode lobe of insertion (horizontal versus verticalmodels), all simulations were performed using the horizontalelectrode insertion, representing placement in the middle lobe.

2) Electric Field Distribution: The electric field distributionis determined according to

∇ · (σ(|E|)∇φ) = 0 (1)

where σ is the electrical conductivity of the tissue, E is theelectric field in V/m, and φ is the electrical potential. Tissue–electrode boundaries were defined as φ = Vo or 0. The re-maining boundaries were considered electrically insulating,

dφ/dn = 0, since the kidneys were isolated from the surround-ing mesenchymal tissue during the procedures.

3) Electrical Conductivity Behavior and Functions: Themodel was calibrated to the effective renal bulk tissue con-ductivity and response to the electric pulses, including one withhomogeneous static conductivity (σ0), and two dynamic modelsthat accounted for EP-based conductivity changes. The dynamicmodels relate baseline and maximum conductivities in a con-tinuous function based on local electric field exposure that canbe matched to overall bulk tissue properties. The baseline con-ductivity, σ0 , was determined by matching simulated prepulseelectrical current from the static conductivity model with theexperimental data, where field strength should be below theEP threshold of cells in the tissue. The maximum conductivity,σmax , occurs when the cellular membranes no longer restrict theextent of interstitial electrolyte mobility. The statistical modelin [38] was used to predict σmax from previously characterizedtissue prepulse response to σ0 , using parametric analyses ofmodels and electrical data from other tissues. The σ0 and σmaxvalues were determined uniquely for each model conductivitycondition. The conductivity values were calibrated as functionsof the electric field to provide overall bulk tissue effective con-ductivities that match the measured electrical currents from theexperimental pulse protocols, producing model-specific param-eters for the three electrical conductivity conditions.

The σ0 and σmax values provide the parameters to defineelectric field-dependent conductivity, σ(|E|), of renal tissue invivo. One model assumed a linear relationship, σL (E), whereσ0 and σmax were related by a linear relationship over the rangebetween 200 and 2000 V/cm. The second dynamic conductivitysimulation used electrical conductivity conditions of an asym-metrical sigmoid Gompertz curve σG (E) [30]:

σG (|E|) = σ0 + (σmax − σ0) · exp[−A · exp(−B · E)] (2)

where A and B are unitless coefficients that vary with pulselength, t (s). This function was fit using curve parameters for a100 μs long pulse, where A = 3.053 and B = 0.00233 [30].

4) Lethal EIRE Threshold Determination: The finite-element model simulated the electric field distribution to de-termine EIRE by correlating numerical results with the in vivolesion height and width dimensions. The electric field distri-bution at the midpoint length of the electrodes determined theelectric field magnitude that matched experimental lesion di-mensions. This was performed to determine which model bestmatched all three IRE ablation protocols. Tests of statisticalsignificance between ablation protocols, as well as calibratedelectric field threshold, were calculated with Student’s t-test andANOVA using Microsoft Excel (Microsoft, Redmond, WA).

III. RESULTS

A. In Vivo Experiments

1) Electrical Current: All animals survived the proceduresuntil euthanasia. Prepulse and experimental electric currents areshown in Table II, where experimental trials displayed typi-cal waveforms to those in [39] [see Fig. 1(c) and (d)]. Therewas no significant difference in experimental currents between


TABLE IIEXPERIMENTAL ELECTRIC CURRENTS TO CALIBRATE MODEL PROPERTIES

Set Separation(cm)

Averagedelivered

voltage (V)

Pulsenumber

Averageelectric

current (A)

Pre 1 1 48 (0.7) 1–2 0.258(0.036)

Pre 2 1.5 73 (1.5) 1–2 0.343(0.050)

1 1 1258 (4.1) 1–10 10.4 (1.7)1257 (2.5) 40–50 11.1 (1.1)1257 (2.9) 90–100 11.0 (1.7)

Average 10.832 2 1760 (9.9) 1–10 20.6 (3.2)

1751 (15.8) 40–50 23.7 (5.1)1753 (12.7) 90–100 23.6 (3.8)

Average 22.633 1.5 2262 (7.6) 1–10 23.6 (1.5)

2262 (7.4) 40–50 24.3 (3.3)2263 (6.2) 90–100 25.4 (4.5)

Average 24.43

∗Dimensions given as “average (standard deviation).”

TABLE IIIIn Vivo EXPERIMENTAL LESION DIMENSIONS

Set Separation(cm)

Voltage–distance

ratio (V/cm)

Width(mm)∗

Height(mm)∗

Area (cm2)∗

1 1 1250 15.9 (1.3) 7.6 (0.60) 1.0 (0.13)2 1 1750 19.0 (1.0) 12.0 (0.55) 1.7 (0.12)3 1.5 1500 22.6 (1.9) 12.1 (0.92) 2.1 (0.33)


horizontal (middle lobe) and vertical (outer lobes) electrode in-sertion approaches. Average current from the first ten pulsesfor each protocol was used to determine σmax in the numeri-cal models. There was no statistically significant relationshipbetween pulse number and current, suggesting that cumulativetemperature rise from the pulses did not contribute to bulk tis-sue electrical properties for these pulsing protocols, and thusthermal effects were excluded from the numerical simulations.

2) Lesion Dimensions: Lesions were clearly visible as anellipsoidal hemorrhagic region centered between the electrodeplacements. The TTC dye trials (not shown) turned viable tis-sue outside the lesion bright red, which was absent within thehemorrhagic zone. Lesion size did not correlate with lobe. Le-sions were primarily contained within the renal cortex. Previ-ous in vivo porcine renal experiments show that ablation zoneswill preferentially be drawn toward the renal medulla [40]. Al-though some lesion portions from our trials extended in thedepth dimension into the renal medulla, the extent of medullainteraction did not notably change lesion cross-sectional dimen-sions. Therefore, model tissue properties did not consider renalmedulla interaction. Statistically significant (p < 0.05) differ-ences in lesion width and area are observed between all setups[see Table III and Fig. 2], while height was similar for 1 cm1750 V/cm and 1.5 cm 1500 V/cm.

3) Histology: IRE lesion histology in renal cortex [seeFig. 3] showed a central zone of complete cellular necrosis

Fig. 2. Ablation lesion dimensions from IRE in canine renal tissue.∗p � 0.05,

∗∗p � 0.01, ∗∗∗p < 0.001.

surrounded by a peripheral transition zone, with differential vi-ability based on tissue structure and type, before reaching thecomplete ablation boundary. The transition zone showed addi-tional viable structures in cortex tissue with increasing distancefrom the electrodes. The first to survive the pulses were arteri-oles and other vasculature, followed by glomeruli, distal convo-luted tubules (DCTs), and finally proximal convoluted tubules(PCTs). Renal medulla and the collecting system are composedmainly of vascular bundles and collecting ducts, making themmore homogeneous than cortex. The completely necrotic tocompletely unaffected transition was thus much sharper in theseregions than the cortex.

B. Numerical Modeling

1) Kidney Segmentation and 3-D Reconstruction: Fig. 4(a)shows the canine kidney 3-D MRI volumetric mesh with themidpoint plane illustrating where electric field distributionswere calibrated. Fig. 4(b)–(d) shows electric field along twoperpendicular lines along the height or width from the center ofthe midpoint plane [see Fig. 4(a), red lines], used to calibrateEIRE . Max electric field difference between the center versusouter lobe orientation was


Fig. 3. Histology of renal tissue 6 h post-IRE. (a)–(f) Cortex and (g, h) medulla tissue. (a) Normal renal cortex showing viable glomeruli (circled), proximalconvoluted tubules (PCT, arrows), and distal convoluted tubules (DCT, arrowheads), 100×. (b) Complete cellular necrosis within majority of lesion zone, 100×.(c, d) Lesion region immediately adjacent to electrode insertion tract (∗) showing complete cellular necrosis and pyknotic nuclei; c: 40×, d: 100×. Some structuralECM remains discernible, including PCT (arrow) and DCT (arrowhead). (e) Viable arteriole (arrow) and PCT (arrowhead) with adjacent necrotic tubules, 100×.(f) Early transition zone showing viable glomeruli (circled) amongst necrotic tubule, 100×. (g) Regular medulla architecture is observed 4 mm from electrodeinsertion tract, 100×. (h) Necrotic medulla immediately adjacent to electrode insertion tract, 100×.

Fig. 4. Numerical simulation output. (a) MRI-based volumetric mesh withelectrodes inserted in horizontal position. The electric field calculated fromthe computational models along the width and height dimensions at thecross-sectional plane in the center of the electrode exposure as depicted in(a) for the (b) 1.0 cm separation, 1250 V, (c) 1.0 cm separation, 1750 V, and(d) 1.5 cm separation, 2250 V experimental IRE cases for the horizontal andvertical experimental approaches using σG (|E |).

along the x–z [see Fig. 5(a), (c), and (e)] and x–y [see Fig. 5(b),(d), and (f)] cross-sectional planes from the three experimen-tal setups, demonstrating the larger electric field coverage witha higher applied voltage over the same separation distance, aswell as the effect of voltage-to-distance ratio.

The electric field distributions from all three modeling tech-niques were correlated with the experimental lesion height andwidth dimensions (see Table III) to provide EIRE values foreach experimental setup (see Table VI). The Gompertz dy-namic electric conductivity model has the least EIRE variabilityfor both dimensions and all three experimental setups; withEIRE ranges of 195 V/cm (485–680 V/cm), versus 324 V/cm(373–697 V/cm) for the static conductivity and 223 V/cm (563–707 V/cm) for the linear dynamic conductivity models. Mean

TABLE IVCALIBRATED BASELINE AND MAXIMUM RENAL ELECTRICAL CONDUCTIVITY

Setup Separation(cm)

Voltage–distance

ratio(V/cm)

σ0(S/m)

σL m a x(S/m)

σL m a x/σ0 σG m a x(S/m)

σG m a x/σ0

1 1 1250 0.365 0.942 2.58 0.763 2.092 1 1750 0.365 1.355 3.71 1.150 3.153 1.5 1500 0.341 1.287 3.77 1.050 3.08

Average 0.353 1.195 3.39 0.988 2.80

TABLE VNUMERICALLY SIMULATED ELECTRICAL CURRENTS

Set Separation(cm)

Voltage–distance

ratio (V/cm)

In Vivocurrent

(A)

Model current, A∗

σo(S/m)

σL (|E |)(S/m)

σG (|E |)(S/m)

1 1 1250 10.4 5.7 (45) 12.8 (23) 13.3 (28)2 1 1750 20.6 8.0 (61) 20.4 (1.0) 19.8 (3.9)3 1.5 1500 23.6 9.6 (59) 23.0 (2.5) 22.9 (3.0)

TABLE VIIRE ELECTRIC FIELD THRESHOLDS AT VARYING CONDUCTIVITY CONDITIONS

Conductivityfunction

Set Voltage–distance

ratio (V/cm)

Area E IR E(V/cm)

HeightE IR E

∗

(V/cm)

WidthE IR E

∗

(V/cm)

EH /EW

σo 1 1250 545 445 (26) 600 (172) 0.742 1750 465 373 (22) 453 (87) 0.823 1500 565 440 (29) 697 (264) 0.63

σL(|E|) 1 1250 675 671 (33) 605 (116) 1.112 1750 615 598 (31) 529 (71) 1.133 1500 750 669 (43) 736 (178) 0.91

σG( |E| ) 1 1250 620 605 (29) 575 (117) 1.052 1750 557 526 (27) 485 (67) 1.083 1500 665 582 (37) 680 (184) 0.86



Fig. 5. Numerical model electric field isocontours (V/cm). Calculated electricfield isocontours from the computational simulations in the (a, c, e) x–z planeand the corresponding cross-section in the (b, d, f) x–y plane along the mid-point of the exposed electrode length for the (a, b) 1.0 cm separation, 1250 V;(c, d) 1.0 cm separation, 1750 V; and (e, f) 1.5 cm separation, 2250 V IRE pulseconditions.

and standard deviations for EIRE shows the smallest standarddeviation in the Gompertz model (575 ± 67 V/cm) versus static(501 ± 121 V/cm) or linear dynamic (638 ± 76 V/cm) models.The Gompertz dynamic function electric field distribution bestapproximates lesion shapes, with average height-to-width as-pect ratios (EIRE ,H /EIRE ,W = X) over the experimental setupsof 1.00 ± 0.12 (mean ± SD). A suitable ratio of 1.04 ± 0.13was found for the linear dynamic model, but the static modelratio of lesion shapes was 0.73 ± 0.10, indicating the staticmodels consistently underestimate electric field height dimen-sions. Although numerical model calibrations of lesion areasdo not account for shape variability, the corresponding elec-tric field thresholds were 614 ± 54 V/cm for the σG model,525 ± 53 V/cm for σ0 model, and 680 ± 68 V/cm for the σLmodel.

Improved prediction precision is evident from experimentallesions [see Fig. 6(a), (d), and (g)] superimposed with electricfield isocontours from σ0 and σG (|E|) models [see Fig. 6]. Theσ0 model [see Fig. 6(c), (f), and (i)] required a broader electricfield range to capture lesion shape due to poor height-to-widthelectric field distribution shape relative to the σG (|E|) model[see Fig. 6(b), (e), and (h)].

These results show a model using an asymmetrical sigmoiddynamic conductivity function, σG , with σ0 = 0.353 S/m andσmax = 0.988 S/m offer the best simulation of IRE renal abla-tions. The σL model also provided good consistency, suggestingthat linear dynamic models are suitable for predicting therapeu-tic EP volumes. At 100 pulses, each 100 μs long, the ablationvolume corresponds to the σG modeled 575 ± 67 V/cm electricfield using clinically relevant geometries and voltages.

Fig. 6. Numerical simulation calibrations. (a, d, g) Representative renal abla-tion cross sections within the renal cortex with clearly delineated hemorrhagicregions, which have been overlaid with (b, e, h) asymetrical Gompertz dynamicand (c, f, i) static electrical conductivity numerical simulation results (equalscaling). (a–c) 1 cm separation, 1250 V, (d–f) 1 cm separation, 1750 V, and(g–i) 1.5 cm separation 2250 V. The numerical simulation contours, given inV/cm, include the calibrated EIRE range (± SD) (highlighted). From these, itcan be seen that the dynamic Gompertz function model has a tighter range andbetter general shape to match true experimental in vivo lesions.

IV. DISCUSSION

This study correlates numerical models with in vivo IRE re-nal ablations to determine the relevant electrical properties andlethal electric field threshold for the pulsing protocol used. Exvivo porcine core samples generated a σG (|E|) function be-tween initial and maximum electrical conductivities [30]. Here,the improved accuracy of the σG (|E|) function is shown in vivo,where it generates the best electric field distribution to matchablations out of the three conductivity styles examined, as in-dicated by the smaller standard deviations of predicted ablationshapes. The linear dynamic conductivity model produced nearlyequal quality predictions, and both approaches offer significantimprovement over static conductivity models.

Dynamic conductivity models preferentially expand electricfield in the height dimension. Improving EIRE calibrations tomatch true lesion shapes via the EH /EW ratio is critical, asablation height and width dimensions are often used as the guid-ing criterion when determining appropriate electrode placementarrays for complete targeted volume coverage. Lesion depthis controllable by changing exposure lengths or utilizing thepulse-withdraw-pulse technique. Calibrating lesions to over-all ablation volumes or cross-sectional areas is inappropriatefor determining optimal modeling techniques, as they offer noshape characteristics. Following confirmation of an appropriatemodeling technique, additional ablation metrics such as areaand volume can refine EIRE accuracy.

Lesion growth and resolution is a complex process that mayevolve over days to weeks, with the time of maximum dimen-sions difficult to determine, and possibly varying with tissue


conditions such as tissue type, vascularity, and immunologicalstatus. Therefore, the 6 h post-IRE period for lesion develop-ment was selected to capture clearly delineated lesions, shownby 2 h post-IRE, with additional development and apoptosismarkers found by 6 h [35], [36]. Adjuvant apoptotic, immuno-logical, and ischemic reactive components may contribute tolarger ablation volumes over several days to weeks post-IRE[11], [41], which would lower calibrated absolute EIRE , but theextent of which remains to be defined and is a topic of furtherinquiry for ongoing studies.

Although strong radiologic-pathologic correlation has beenshown at 2 h post-IRE [35], pathologically confirmed lesionswere used in this study to calibrate the numerical simulationsrather than imaging. This was due to the complex and dynamicresponse of tissue to IRE, including edema and vascular occlu-sion, which can complicate viability examination at the 6 h forlesion evolution here. It should be noted that formalin fixationprior to histologically confirming ablated regions will shrinkthe sectioned tissues, and thus the measured lesions, increas-ing calibrated EIRE . This may be responsible for the smallerhistological ablations versus numerical predictions in [25], aswell as CT results at 0 and 1 day post-IRE. Future treatmentsmay reduce this effect by using rapid frozen tissue samples todetermine cellular effects for lesion evaluation, although theimproved lesion shape matching of the dynamic conductivitymodels should persist in any context.

Valuable examination into the ability for existing treatmentplan algorithms to predict ablation outcomes in porcine renaltissue was made in [25]. Static conductivity models simulatedsix experimental IRE protocols using previously calibrated liverEIRE thresholds, which seem to best match the 24 h post-IRECT scans of the central nonenhancing core. This region is de-scribed to likely indicate the region of complete ablation ratherthan the hyperemic rim, which may match the pathological tran-sition zone, and was considerably larger than predictions. Allablation examinations were larger than the histologically con-firmed sections, possibly from formalin-induced tissue contrac-tion. Their study offers valuable experimental data to attainadditional insight when comparing numerical simulations within vivo ablations. This includes evaluation using dynamic con-ductivity models. Additionally, new kidney-specific numericalmodel calibrations may further improve modeling predictive ca-pacity of their ablations. Finally, where our study examines asingle set of secondary pulse parameters (100 μs, 100 pulses) tostandardize an effective EIRE for these specific conditions, theirstudy used several secondary pulse parameters, thus further ex-amination with newly calibrated thresholds would offer insightfor how these parameters influence effective EIRE thresholds.

While it is established that temperature changes will affect tis-sue electrical conductivity [42], [43], their effects were excludedfrom the numerical models. This is because the electric currentmeasurements showed no notable relationship between currentand pulse number, where higher pulse numbers would reflectcumulative temperature rise. Although the results are consistentwith the in vivo findings in [40], the absence of current rise withpulse number is contrary to the findings in [30], where a correla-tion could be attributed to the predicted rise of 1–3%·°C−1 from

the literature [42]. The study in [30] used ex vivo core sampleshomogeneously exposed to electric fields via plate electrodes,without heat conduction to surrounding tissues or blood perfu-sion heat sink, contrasting with the in vivo trials here. Thermaleffects exude less influence on tissue conductivity and electricfield redistribution than EP [41], with little effect on EIRE . Fu-ture work may consider thermal effects by adding a temperatureterm to the dynamic electrical conductivity function [32].

The Gompertz conductivity function produced electric fieldsthat best matched the in vivo lesion shapes, determined byEH /EW ratios. The linear dynamic conductivity function wasequivalent within statistical significance. Previous literatureshows dynamic conductivity equations, including step, linear,sigmoid, and exponential functions, improve electric currentand affected volume size predictions [28], [29], [33], [37],[44]. Model EIRE standard deviations here were primarily at-tributable to experimental variability.

This study used three electrode geometry and voltage com-binations to derive renal tissue electrical properties and EIRE .This permits treatment planning under varying electrode geome-tries and voltages, such as those described theoretically in [45]–[47], enabling patient-specific simulations to plan treatments.Secondary pulse parameters (pulse length, repetition rate, andnumber of pulses) will alter susceptibility to IRE and cumula-tive conductivity behavior of the tissue, as shown in vitro [48]and in vivo [41]. Therefore, the EIRE determined here will varyif these parameters are significantly changed from the 100 μs,100 pulses at 1 pulse/s, which is within typical clinical andexperimental ranges [49].

Although electrical parameters may vary for other tissues, thisstudy supports using dynamic conductivity simulations, partic-ularly an asymmetrical sigmoid function, to provide accuratelesion dimension approximations. While explicitly determiningrelevant conductivity parameters for each tissue for IRE remainsimportant, the β-dispersion hypothesis in [30] may approximatenecessary parameters in the absence of in vivo data. This hy-pothesis predicts conductivity parameters by relating saturatedEP tissue response with tissue behavior in the MHz AC fre-quency range, which also negates cell membrane influence ontissue conductivity.

Histology offers interesting considerations here. IRE is notedto produce sharp delineations between dead and unaffected cellsin healthy liver [6], a relatively isotropic and homogeneousorgan. However, a 1–3 mm transition zone is noted beyond thewholly necrotic region for renal cortex cells, which appearsdependent on the cell and microstructure. Progressively furtherfrom the electrodes, the first surviving structures were bloodvessels and glomeruli, followed by DCTs, and finally PCTs.This observation is consistent with the findings in porcine renalablations [25].

Transition zone structure viability correlation with electricfield suggests possible EIRE cell dependence and the potentialinfluence of secondary cell death mechanisms, including is-chemia and apoptosis. Higher PCT energy demands comparedto other renal structures may make them less resilient to pulse-mediated vascular occlusion ischemia or membrane integrity-induced stresses. Because rapidly dividing cancer cells typically


have high energy demands, it is possible they may be killed atlower pulse protocol strengths than healthy, fully differentiatedcells more tolerant of EP-induced stresses. This prospect re-mains speculative, but supports findings in [50], where IREkilled sarcoma cells while adjacent muscle cells survived. Therelatively homogeneous distribution of cellular constituents inrenal medulla affected from our trials did not exhibit this tran-sition zone.

This work used healthy renal tissue to derive EIRE and vali-date dynamic conductivity modeling techniques. Further in vivowork should determine IRE efficacy and tissue properties in re-nal tumors. The results may be combined to optimize treatmentplans that can ensure complete destruction of the tumorous tis-sue while understanding the extent of damage to healthy tissue.Such further refined simulations may also account for other lo-calized tissue heterogeneities such as the distribution of majorvasculature or the collecting system.

Short to medium-term preclinical studies investigatingmacro- and microscopic effects of IRE in healthy renal tis-sue show complete cell death of glomerular and tubule cor-tex structures while sparing major vasculature. Initial heteroge-neous urothelium effects were noted, but did not damage theurinary system 28 days post-IRE due to regeneration of theurinary system and preservation of urothelial basement mem-branes, with no urinoma evidence [21], [22]. Renal lesions 24and 36 h post-IRE showed a transition zone between necroticand normal tissue with evidence of tubule degeneration andcomplete cell necrosis beyond lesions [51]. Two to four week ab-lations became contracted scars, with cortex structures replacedby fibrous tissue. There was evidence within ablations of mor-phologically viable and regenerating tubules, as well as intactECM and regenerative pelvic epithelium. Urinary tract effectsshowed protection of the urine-collecting system with regen-erated urothelial tissue [24]. In addition, angiographic contrastevaluation on ex vivo perfused porcine kidneys showed no acutecollateral vascular damage [23]. Early clinical investigationsdemonstrate the safety profile of IRE renal ablation performedimmediately prior to resection [52], and have reported promis-ing early therapeutic potential for IRE as a stand-alone modalityfor renal tumor treatment [14].

V. CONCLUSION

This study performed in vivo IRE renal ablations to cali-brate numerical modeling techniques employing three distinctelectrical conductivity conditions. Varying electrode geometryand pulse parameters demonstrate the ability to customize ab-lation regions for a given pulse protocol. Histology showed afully necrotic ablation zone within a peripheral transition zone,demonstrating heterogeneous lethality dependent on structural,tissue, and cell variety. Computational models simulating theexperiments determined in vivo renal electrical properties forEP-based treatment planning to determine a kidney-specific ef-fective lethal electric field threshold. An asymmetrical sigmoiddynamic electrical conductivity function provided more accu-rate and precise predictions of electrical currents and experimen-tal lesion shape than static models, with linear dynamic modelsnearly equally suitable. These dynamic conductivity functions

and the lethal electric field threshold may be used in numeri-cal models to aid renal IRE treatment plans and predict healthyrenal parenchyma damage for a protocol employing similar pa-rameters to the 100 pulses, 100 μs long pulses used in our invivo procedures.

ACKNOWLEDGMENT

The authors thank R. Ou, C. Egan, and the AMREP ani-mal services staff for their assistance with the experimentalprocedures.

REFERENCES

[1] W. Kassouf, A. G. Aprikian, M. Laplante, and S. Tanguay, “Natural historyof renal masses followed expectantly,” J. Urol., vol. 171, pp. 111–113,2004.

[2] S. Ginzburg, R. Uzzo, and A. Kutikov, “The role of minimally invasivesurgery in multifocal renal cell carcinoma,” Curr. Urol. Rep., vol. 13,pp. 202–210, 2012.

[3] A. M. Venkatesan, B. J. Wood, and D. A. Gervais, “Percutaneous ablationin the kidney,” Radiology, vol. 261, pp. 375–391, Nov. 1, 2011.

[4] R. N. Uppot, S. G. Silverman, R. J. Zagoria, K. Tuncali, D. D. Childs,and D. A. Gervais, “Imaging-guided percutaneous ablation of renal cellcarcinoma: A primer of how we do it,” Amer. J. Roentgenol., vol. 192,pp. 1558–1570, Jun. 1, 2009.

[5] A. Hines-Peralta and S. N. Goldberg, “Review of radiofrequency ablationfor renal cell carcinoma,” Clin. Cancer Res., vol. 10, pp. 6328S–6334S,Sep. 15, 2004.

[6] J. F. Edd, L. Horowitz, R. V. Davalos, L. M. Mir, and B. Rubinsky,“In vivo results of a new focal tissue ablation technique: Irreversibleelectroporation,” IEEE Trans. Biomed. Eng., vol. 53, no. 7, pp. 1409–1415, Jul. 2006.

[7] R. Davalos, L. Mir, and B. Rubinsky, “Tissue ablation with irreversibleelectroporation,” Ann. Biomed. Eng., vol. 33, pp. 223–231, 2005.

[8] J. C. Weaver and R. A. Mintzer, “Decreased bilayer stability due to trans-membrane potentials,” Phys. Lett. A, vol. 86, pp. 57–59, 1981.

[9] G. Onik, P. Mikus, and B. Rubinsky, “Irreversible electroporation: Im-plications for prostate ablation,” Technol. Cancer Res. Treatment, vol. 6,pp. 295–300, 2007.

[10] B. Rubinsky, G. Onik, and P. Mikus, “Irreversible electroporation: A newablation modality—Clinical implications,” Technol. Cancer Res. Treat-ment, vol. 6, pp. 37–48, 2007.

[11] E. W. Lee, C. T. Loh, and S. T. Kee, “Imaging guided percutaneous irre-versible electroporation: Ultrasound and immunohistological correlation,”Tech. Cancer Res. Treatment, vol. 6, pp. 287–293, 2007.

[12] R. E. Neal, II, W. Cheung, H. Kavnoudias, and K. R. Thomson, “Spectrumof imaging and characteristics for liver tumors treated with irreversibleelectroporation,” J. Biomed. Sci. Eng., vol. 5, pp. 813–818, 2012.

[13] R. V. Davalos, B. Rubinsky, and D. M. Otten, “A feasibility study forelectrical impedance tomography as a means to monitor tissue electropo-ration for molecular medicine,” IEEE Trans. Biomed. Eng., vol. 49, no. 4,pp. 400–403, Apr. 2002.

[14] K. R. Thomson, W. Cheung, S. J. Ellis, D. Federman, H. Kavnoudias, D.Loader-Oliver, S. Roberts, P. Evans, C. Ball, and A. Haydon, “Investiga-tion of the safety of irreversible electroporation in humans,” J. VascularIntervention Radiol., vol. 22, pp. 611–621, Mar. 25, 2011.

[15] T. P. Kingham, A. M. Karkar, M. I. D’Angelica, P. J. Allen, R. P. Dematteo,G. I. Getrajdman, C. T. Sofocleous, S. B. Solomon, W. R. Jarnagin, and Y.Fong, “Ablation of perivascular hepatic malignant tumors with irreversibleelectroporation,” J. Amer. Coll. Surg., vol. 215, pp. 379–87, 2012.

[16] G. Narayanan, P. J. Hosein, G. Arora, K. J. Barbery, T. Froud, A. S. Living-stone, D. Franceschi, C. M. Rocha Lima, and J. Yrizarry, “Percutaneousirreversible electroporation for downstaging and control of unresectablepancreatic adenocarcinoma,” J. Vascular Intervention Radiol., vol. 23, pp.1613–1621, 2012.

[17] W. Cheung, H. Kavnoudias, S. Roberts, B. Szkandera, W. Kemp, and K.R. Thomson, “Irreversible electroporation for unresectable hepatocellularcarcinoma: Initial experience and review of safety and outcomes,” Technol.Cancer Res. Treatment, vol. 12, pp. 233–241, 2013.

[18] R. Cannon, S. Ellis, D. Hayes, G. Narayanan, and R. C. G. Martin, 2nd.,“Safety and early efficacy of irreversible electroporation for hepatic tumorsin proximity to vital structures,” J. Surg. Oncol., vol. 107, pp. 544–549,2013.


[19] H. J. Scheffer, K. Nielsen, M. C. de Jong, A. A. J. M. van Tilborg, J.M. Vieveen, A. Bouwman, S. Meijer, C. van Kuijk, P. M. van den Tol,and M. R. Meijerink, “Irreversible electroporation for nonthermal tumorablation in the clinical setting: A systematic review of safety and efficacy,”J. Vascular Intervention Radiol., vol. 25, pp. 997–1011, 2014.

[20] R. E. Neal, 2nd, J. L. Millar, H. Kavnoudias, P. Royce, F. Rosenfeldt, A.Pham, R. Smith, R. V. Davalos, and K. R. Thomson, “In vivo character-ization and numerical simulation of prostate properties for non-thermalirreversible electroporation ablation,” Prostate, vol. 17, pp. 458–68, 2014.

[21] C. R. Tracy, W. Kabbani, and J. A. Cadeddu, “Irreversible electroporation(IRE): A novel method for renal tissue ablation,” BJU Int., vol. 107,pp. 1982–1987, 2011.

[22] J. J. Wendler, M. Porsch, S. Hühne, D. Baumunk, P. Buhtz, F. Fischbach,M. Pech, D. Mahnkopf, S. Kropf, A. Roessner, J. Ricke, M. Schostak, andU. B. Liehr, “Short- and mid-term effects of irreversible electroporationon normal renal tissue: An animal model,” Cardiovascular InterventionalRadiol., vol. 36, pp. 512–520, 2013.

[23] J. Wendler, M. Pech, S. Blaschke, M. Porsch, A. Janitzky, M. Ulrich, O.Dudeck, J. Ricke, and U. B. Liehr, “Angiography in the isolated perfusedkidney: Radiological evaluation of vascular protection in tissue ablation bynonthermal irreversible electroporation,” Cardiovascular InterventionalRadiol., vol. 35, pp. 383–390, 2012.

[24] J. J. Wendler, M. Pech, M. Porsch, A. Janitzky, F. Fischbach, P. Buhtz,K. Vogler, S. Huhne, K. Borucki, C. Strang, D. Mahnkopf, J. Ricke, andU. B. Liehr, “Urinary tract effects after multifocal nonthermal irreversibleelectroporation of the kidney: Acute and chronic monitoring by magneticresonance imaging, intravenous urography and urinary cytology,” Cardio-vascular Interventional Radiol., vol. 35, pp. 921–926, 2012.

[25] T. Wimmer, G. Srimathveeravalli, N. Gutta, P. C. Ezell, S. Monette,M. Maybody, J. P. Erinjery, J. C. Durack, J. A. Coleman, and S. B.Solomon, “Planning irreversible electroporation in the porcine kidney:Are numerical simulations reliable for predicting empiric ablation out-comes?” Cardiovascular Interventional Radiol., May 17, 2014.

[26] A. T. Esser, K. C. Smith, T. R. Gowrishankar, and J. C. Weaver, “Towardssolid tumor treatment by irreversible electroporation: Intrinsic redistribu-tion of fields and currents in tissue,” Technol. Cancer Res. Treatment, vol.6, pp. 261–273, 2007.

[27] R. E. Neal, 2nd and R. V. Davalos, “The feasibility of irreversible elec-troporation for the treatment of breast cancer and other heterogeneoussystems,” Ann. Biomed. Eng., vol. 37, pp. 2615–2625, 2009.

[28] D. Sel, D. Cukjati, D. Batiuskaite, T. Slivnik, L. M. Mir, and D. Miklavcic,“Sequential finite element model of tissue electropermeabilization,” IEEETrans. Biomed. Eng., vol. 52, no. 5, pp. 816–827, May 2005.

[29] N. Pavselj, Z. Bregar, D. Cukjati, D. Batiuskaite, L. M. Mir, andD. Miklavcic, “The course of tissue permeabilization studied on a math-ematical model of a subcutaneous tumor in small animals,” IEEE Trans.Biomed. Eng., vol. 52, no. 8, pp. 1373–1381, Aug. 2005.

[30] R. E. Neal, II, P. A. Garcia, J. L. Robertson, and R. V. Davalos, “Ex-perimental characterization and numerical modeling of tissue electricalconductivity during pulsed electric fields for irreversible electroporationtreatment planning,” IEEE Trans. Biomed. Eng., vol. 59, no. 4, pp. 1076–1085, Apr. 2012.

[31] A. Ivorra and B. Rubinsky, “In vivo electrical impedance measurementsduring and after electroporation of rat liver,” Bioelectrochemistry, vol. 70,pp. 287–295, May 2007.

[32] P. Garcia, J. Rossmeisl, R. Neal, T. Ellis, J. Olson, N. Henao-Guerrero,J. Robertson, and R. V. Davalos, “Intracranial nonthermal irreversibleelectroporation: In vivo analysis,” J. Membr. Biol., vol. 236, pp. 127–136,2010.

[33] S. Corovic, I. Lackovic, P. Sustaric, T. Sustar, T. Rodic, and D. Miklavcic,“Modeling of electric field distribution in tissues during electroporation,”Biomed. Eng. Online, vol. 12, pp. 1–27, 2013.

[34] R. E. Neal II, P. A. Garcia, H. Kavnoudias, F. Rosenfeldt, C. Mclean,V. Earl, J. Bergman, R. V. Davalos, and K. R. Thomson “Simulation ofin vivo irreversible electroporation renal ablations,” presented at the 6thEuropean Conference of the International Federation for Medical andBiological Engineering, Dubrovnik, Croatia, pp. 813–816, 2014.

[35] L. Appelbaum, E. Ben-David, J. Sosna, Y. Nissenbaum, and S. N. Gold-berg, “US findings after irreversible electroporation ablation: Radiologic-pathologic correlation,” Radiology, vol. 262, pp. 117–125, 2012.

[36] B. Al-Sakere, F. Andre, C. Bernat, E. Connault, P. Opolon, R. V. Davalos,B. Rubinsky, and L. M. Mir, “Tumor ablation with irreversible electropo-ration,” PLoS One, vol. 2, p. e1135, 2007.

[37] M. Pavlin, M. Kanduser, M. Rebersek, G. Pucihar, F. X. Hart, R. Magjare-vic, and D. Miklavcic, “Effect of cell electroporation on the conductivityof a cell suspension,” Biophys. J., vol. 88, pp. 4378–4390, 2005.

[38] P. A. Garcia, C. B. Arena, and R. V. Davalos, “Towards a predictivemodel of electroporation-based therapies using pre-pulse electrical mea-surements,” in Proc. Conf. IEEE Eng. Med. Biol. Soc., vol. 2012, 2012,pp. 2575–2578.

[39] D. Cukjati, D. Batiuskaite, F. André, D. Miklavčič, and L. M. Mir, “Realtime electroporation control for accurate and safe in vivo non-viral genetherapy,” Bioelectrochemistry, vol. 70, pp. 501–507, 2007.

[40] E. Ben-David, M. Ahmed, M. Faroja, M. Moussa, A. Wandel, J. Sosna,L. Appelbaum, I. Nissenbaum, and S. N. Goldberg, “Irreversible electro-poration: treatment effect is susceptible to local environment and tissueproperties,” Radiology, vol. 269, pp. 738–747, 2013.

[41] Z. Qin, J. Jiang, G. Long, B. Lindgren, and J. C. Bischof, “Irreversibleelectroporation: An in vivo study with dorsal skin fold chamber,” Ann.Biomed. Eng., vol. 41, pp. 619–629, 2013.

[42] F. A. Duck, Physical Properties of Tissue: A Comprehensive ReferenceBook. New York, NY, USA: Academic, 1990.

[43] U. Pliquett, “Joule heating during solid tissue electroporation,” Med. Biol.Eng. Comput., vol. 41, pp. 215–219, 2003.

[44] A. Ivorra, L. M. Mir, and B. Rubinsky, “Electric field redistribution dueto conductivity changes during tissue electroporation: Experiments witha simple vegetal model,” presented at the World Congress on MedicalPhysics and Biomedical Engineering, Munich, Germany, vol. 25/13, 2009,pp. 59–62.

[45] J. F. Edd and R. V. Davalos, “Mathematical modeling of irreversibleelectroporation for treatment planning,” Technol. Cancer Res. Treatment,vol. 6, pp. 275–286, 2007.

[46] D. Sel, A. M. Lebar, and D. Miklavcic, “Feasibility of employing model-based optimization of pulse amplitude and electrode distance for effectivetumor electropermeabilization,” IEEE Trans. Biomed. Eng., vol. 54, no. 5,pp. 773–781, May 2007.

[47] A. Županič, S. Čorović, and D. Miklavčič, “Optimization of electrodeposition and electric pulse amplitude in electrochemotherapy,” Radiol.Oncol., vol. 42, pp. 93–101, 2008.

[48] G. Pucihar, J. Krmelj, M. Rebersek, T. B. Napotnik, and D. Miklavcic,“Equivalent pulse parameters for electroporation,” IEEE Trans. Biomed.Eng., vol. 58, no. 11, pp. 3279–3288, Nov. 2011.

[49] E. Ben-David, L. Appelbaum, J. Sosna, I. Nissenbaum, and S. N. Goldberg,“Characterization of irreversible electroporation ablation in in vivo porcineliver,” Amer. J. Roentgenol., vol. 198, pp. W62–W68, 2012.

[50] R. E. Neal, 2nd, J. H. Rossmeisl, Jr., P. A. Garcia, O. I. Lanz, N. Henao-Guerrero, and R. V. Davalos, “Successful treatment of a large soft tissuesarcoma with irreversible electroporation,” J. Clin. Oncol., vol. 14, pp.e372–7, 2011.

[51] A. Deodhar, S. Monette, G. W. Single, Jr., W. C. Hamilton, Jr., R. Thornton,M. Maybody, J. A. Coleman, and S. B. Solomon, “Renal tissue ablationwith irreversible electroporation: Preliminary results in a porcine model,”Urology, vol. 77, pp. 754–760, 2011.

[52] M. Pech, A. Janitzky, J. J. Wendler, C. Strang, S. Blaschke, O. Dudecket al., “Irreversible electroporation of renal cell carcinoma: A first-in-manphase I clinical study,” Cardiovascular Interventional Radiol., vol. 34,pp. 132–138, 2011.

Authors’, photographs and biographies not available at the time of publication.

/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 150 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages false /GrayImageDownsampleType /Bicubic /GrayImageResolution 1200 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.00083 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages false /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages false /MonoImageDownsampleType /Bicubic /MonoImageResolution 1600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.00063 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False

/CreateJDFFile false /Description >>> setdistillerparams> setpagedevice

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, … · 562 IEEE TRANSACTIONS ON BIOMEDICAL...

Documents

Transcript of IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, … · 562 IEEE TRANSACTIONS ON BIOMEDICAL...